We feel that the problem addressed by the authors is extremely important and we agree very much with the general philosophy of their work. Posterior model probabilities can be a very effective tool in communicating model uncertainty to clients. In our exp (1994)

by M A CLYDE, H DeSIMONE, G PARMIGIANI
Venue:MC and George and McCulloch's 1994 integrated SSVS in Clyde and Parmigiani